Syntactic Simplification for Improving Content Selection in Multi-Document Summarization

Syntactic Simplification for Improving Content Selection in Multi-Document Summarization Advaith Siddharthan, Ani Nenkova and Kathleen McKeown Columbia University Computer Science Department ������ �  ��� � � � @ �  �  �� "! � Abstract In this paper, we explore the use of automatic syntactic simplification for improving content selection in multi-document summarization. In particular, we show how simplifying parenthet- icals by removing relative clauses and apposi- tives results in improved sentence clustering, by forcing clustering based on central rather than background information. We argue that the in- clusion of parenthetical information in a sum- mary is a reference-generationtask rather than a content-selection one, and implement a baseline reference rewriting module. We perform our evaluations on the test sets from the 2003 and 2004 Document Understanding Conference and report that simplifying parentheticals results in significant improvement on the automated eval- uation metric Rouge. 1 Introduction Syntactic simplification is an NLP task, the goal of which is to rewrite sentences to reduce their gram- matical complexity while preserving their meaning and information content. Text simplification is a useful task for varied reasons. Chandrasekar et al. (1996) viewed text simplification as a preprocess- ing tool to improve the performance of their parser. The PSET project (Carroll et al., 1999), on the other hand, focused its research on simplifying newspaper text for aphasics, who have trouble with long sen- tences and complicated grammatical constructs. We have previously (Siddharthan, 2002; Siddharthan, 2003) developed a shallow and robust syntactic sim- plification system for news reports, that simplifies relative clauses, apposition and conjunction. In this paper, we explore the use of syntactic simplification in multi-document summarization. 1.1 Sentence Shortening for Summarization It is interesting to survey the literature in sentence shortening, a task related to syntactic simplification. Grefenstette (1998) proposed the use of sentence shortening to generate telegraphic texts that would help a blind reader (with a text-to-speech software) skim a page in a manner similar to sighted readers. He provided eight levels of telegraphic reduction. The first (the most drastic) generated a stream of all the proper nouns in the text. The second gen- erated all nouns in subject or object position. The third, in addition, included the head verbs. The least drastic reduction generated all subjects, head verbs, objects, subclauses and prepositions and dependent noun heads. Reproducing from an example in his paper, the sentence: Former Democratic National Committee fi- nance director Richard Sullivan faced more pointed questioning from Republicans during his second day on the witness stand in the Senate’s fund-raising investigation. got shortened (with different levels of reduction) to: # Richard Sullivan Republicans Senate. # Richard Sullivan faced pointed questioning. # Richard Sullivan faced pointed questioning from Republicans during day on stand in Senate fund- raising investigation. Grefenstette (1998) provided a rule based ap- proach to telegraphic reduction of the kind illus- trated above. Since then, Jing (2000), Riezler et al. (2003) and Knight and Marcu (2000) have ex- plored statistical models for sentence shortening that, in addition, aim at …

Content truncated to 500 words